Performance under pressure: What four CRE leaders agree on, and why it starts with data

The market environment entering 2026 is something most CRE professionals have learned to describe carefully. It’s not currently in crisis, but a full recovery hasn’t materialized yet either. Instead, we are sitting somewhere in between the two ends of that spectrum, with capital re-entering selectively, transaction activity picking back up, and macroeconomic noise running constantly in the background.

We recently sat down with four leaders from across the CRE ecosystem to unpack what performance and risk management actually look like in this environment. It was a real, candid discussion between CRE leaders who are close enough to the problems to offer perspectives that hold well beyond the conference room.

What emerged was less a debate than a convergence: different seats in the CRE ecosystem, but many of the same fundamental convictions about what it takes to measure and manage well.

Performance measurement requires a framework, not just data

The pressure to explain performance to LPs and boards has never been higher, and data is a natural starting point. But for Jack Dowd, Senior Consultant at Aon Investment Consulting, data alone doesn’t fully satisfy what his clients are actually asking for. His clients, pension plans, and Taft-Hartley funds, among them, are making multi-decade commitments to sponsors and partners. They want the numbers, but the numbers alone never offer the full picture. "Our clients are also focused on some other, more qualitative measures", Dowd said. This includes senior leadership turnover, legal exposure, and the kind of organizational stability that outlasts any single performance cycle.

Melanie Domres, President at NewTower Trust Company, brings a complementary perspective from inside the portfolio. Her firm oversees the valuations and appreciation returns for more than 300 assets across core fund clients. Her orientation leans pragmatic: the decision-making criteria don't change with the environment. "The decision tree is the same regardless of economic circumstances," she said. Buy, sell, or hold – and if hold, at what cost, toward what execution. The data informs those decisions, but it doesn't replace the underlying framework for making them.

Joshua Rome, Senior Director of Performance and Portfolio Analytics at Nuveen Real Estate, adds the analytical layer. His team sits at the intersection of reporting and decision-making across more than $100 billion of AUM, working from three interconnected data silos — operational, appraisal, and performance — that have to be consistent with each other. When they aren't, the conclusions the model produces are unreliable, regardless of how sophisticated the analysis appears.

"When you have the different data points that you force to be consistent, and you build these models from the most granular data up, you're much less likely to draw false conclusions,” Rome said.

The risk in the data is as real as the risk in the market

One point the panel kept returning to is that the data itself carries risk, independent of the market conditions the data is trying to describe.

Rome identified two distinct categories of this risk, with the first being data quality. Financials audited by a fund's accounting team provide a baseline of reliability, as they'll tie to trial balances and they've been validated. But even audited data has failure modes. Month-to-month reversals. Accounting conventions that differ between teams. Inconsistencies that surface only when you try to compare two funds that were booked by two different teams with two different definitions of the same term. A data governance function goes a long way toward preventing this, but it requires institutional commitment, not just good intentions.

The second risk is comparability; attribution models carry an implicit assumption that you're comparing like with like, but that assumption breaks down more often than most performance teams acknowledge. Consider an asset manager whose actuals beat budget and whose comp set looks favorable, yet who is somehow underperforming their benchmark. The instinct is to question the manager. The diagnosis, more often, is what Rome calls "hidden allocation effect"; a different quality of asset, a different rent level, a position in the market that the benchmark doesn't actually reflect. "You can turn that into an insight instead of a misinterpretation,” he noted. But only if you have enough granularity in your data sets to see what's actually driving the divergence.

Domres looks at this through a practitioner's lens. At NewTower, she spends significant time and energy managing what happens when sales comps and leasing comps get used by third-party appraisers in ways that don't reflect the competitive position of a specific asset. "We parse them," she said of the comps. "We do not cherry-pick them." Cherry-picking produces a number that is easy to justify; you can point to the comps that support it. Parsing produces a number that holds up to scrutiny because it accounts for the full picture, versus just a favorable selection of it.

Consistency in the valuation process is what makes performance comparison meaningful

The data quality discipline, framework-building, and the effort to make appraisal, ops, and performance data consistent are in service of something specific: making it possible for LPs to evaluate managers on what actually differentiates them.

Robby Tandjung, EVP of Valuation Advisory at Altus Group, framed this precisely. When valuation processes are consistent across funds, using the same methodologies, same review standards, same platform rigor, the comparison that Dowd's clients are trying to make becomes meaningful. But without that consistency, the conversation between a fund consultant and a fund manager is polluted by valuation differences that have nothing to do with allocation or selection decisions. "Now they can pick the fund based purely on the manager's style, allocation, and selection, rather than valuation differences,” explained Tandjung.

Tandjung also flagged that benchmarking itself is evolving in ways that raise the bar for everyone. Averaging across an index is no longer sufficient as LPs increasingly ask for performance broken down by cashflow effect versus yield effect, by vintage, by building class, and by submarket. "It's no longer just the averages," he explained, noting that a core fund and a core-plus fund require different benchmarks. An industrial portfolio in the Inland Empire requires a different lens than a diversified industrial benchmark. The implication for valuation and performance teams is that the data infrastructure has to be granular enough to support that level of decomposition, not just today, but as LP expectations continue to evolve.

The pressures Tandjung describes are ones Altus has been building toward directly. Altus’ focus has shifted toward connected workflows and proactive performance insight that gives every stakeholder, from appraisers to portfolio managers, a unified view of the same data instead of reconciling across separate versions of it. The benchmarking capability Altus has built reflects the same logic: comparisons drawn from a national dataset of ARGUS DCF models, each calculated on the same engine, which means the granular decomposition Tandjung describes is anchored in consistent inputs and outputs. Solution features like Summary of Assumptions, which automatically surface every assumption embedded in an asset valuation model in one place, directly address the transparency and auditability demands the panel kept returning to.

Shree Guha, Senior Director, Commercial Strategy & Operations at Altus Group, moderated the panel, and drew the thread together: "Consistency in the process drives transparency." It's a deceptively simple formulation for something that is genuinely hard to achieve and genuinely consequential.

AI accelerates the work, but the judgment requirement doesn't move

The discussion closed on a forward-looking note regarding CRE’s evolving relationship with data and AI. Across the panelists, the same conclusion emerged: AI changes the pace of the work, not the nature of it.

Dowd is using Microsoft Copilot to transcribe manager meetings, which means he can look at the person across the table during a conversation rather than at his screen. This is a real upside, but the limitations remain significant; garbage in, garbage out applies to AI-assisted analysis as readily as it applies to any other model. More output doesn't mean better output, and AI-enabled efficiency doesn't shrink the headcount requirement: it expands the scope of what any given team can cover. "Everyone's just working faster, but that doesn't mean we need fewer people,” Dowd said.

Domres zoomed in on the risk that gets less attention: AI-compromised data entering the workflow. She pointed out that some organizations are pulling direct listing data from property sites, including rents that property managers update daily as market conditions shift. That constant churn made the data technically current but analytically challenging: if the numbers change every day, you can't read a trend from them. "We're entering that same paradigm with AI potentially, if we're not careful," Domres said. When AI-generated data gets treated as ground truth without triangulation, the signal disappears into the noise. Human judgment and market knowledge don't go away in an AI-assisted environment. If anything, their value increases.

Looking ahead, Rome was direct about what still needs to change: the industry is too siloed. Research, transactions, asset management, and performance teams have historically talked past each other, each working from data that doesn't speak to anyone else's. But he's cautiously optimistic that connected, consistent data is starting to change the dynamic. When the numbers tell a coherent story across functions, cross-functional alignment becomes possible in a way it simply never was before.

All four perspectives at the table seemed to arrive at the same conclusion: the firms navigating this moment best are the ones that have done the unglamorous foundational work: clean data, consistent processes, frameworks that hold under scrutiny. They are treating AI as an accelerant to that foundation, not a substitute for it.

Want to be notified of our new and relevant CRE content, articles and events?